Speaker: Dr. Kathleen Li, Assistant Professor of Marketing at McCombs School of Business, University of Texas as Austin.
Topic: Statistical Inference for Average Treatment Effects Estimated by Synthetic Control Methods
Abstract: The synthetic control method (SCM), a powerful tool for estimating average treatment effects (ATE), is increasingly popular in fields such as statistics, economics, and marketing and has been called ``arguably the most important innovation in the evaluation literature in the last fifteen years'' (Athey and Imbens 2016). However, SCM has the main limitation that there is no inference theory: therefore, it is not possible to calculate confidence bounds, exact p-values or conduct hypothesis tests. Existing work mostly uses placebo tests. I derive the inference theory for the synthetic control ATE estimators using projection theory, and show that a properly designed subsampling method can be used to obtain confidence intervals and conduct hypothesis test, whereas the standard bootstrap cannot. A second limitation of SCM is that when there is heterogeneity among control units and treatment unit (which occurs frequently in economic and marketing data settings), the synthetic control version of parallel lines assumption may not hold and it can perform poorly in in-sample fit and out-of-sample predictions. I show through an empirical illustration that a modification to the SCM proposed by Doudchenko and Imbens (2016) allows the method to be more widely applied. Simulations and empirical applications examining the effect of opening a physical showroom by Warby Parker and Bonobos demonstrate the usefulness of the inference theory and modified synthetic control method.
Monday, October 15, 2018 at 3:55pm to 5:00pm
Kelley Engineering Center, Room 1003
110 SW Park Terrace, Corvallis, OR 97331